Combined-cycle power plant owners can significantly increase the bottom line by improving the way they treat operations and maintenance costs in dispatch decisions.
By: Paul Schuster, PowerAdvocate and Steven Taub, Cambridge Energy Research Associates
The profitability of combined-cycle gas turbine plants may be falling short due to the common, suboptimal ways that traders, dispatchers and investors approach gas turbine operation and maintenance (O&M) costs in their operating and investment decisions. A case study of an actual plant revealed an opportunity to increase the asset’s value by 23 percent simply by improving the treatment of O&M costs in dispatch decisions. This suggests that there is hidden value for more sophisticated operators and investors to unearth.
Rise and Fall
In the late 1990s, many power industry professionals viewed combined-cycle plants as the ideal solution for the newly emerging competitive markets. When low natural gas prices met growing electricity demand, a large amount of investment ($125 to $150 billion between 1998 and 2005) quickly funneled into these assets. Manufacturers couldn’t produce turbines fast enough; everyone wanted one. Turn it on, let it run, and watch the profits roll in.
And then the market collapsed. Natural gas prices soared, spark spreads vanished, and capacity factors plummeted. In response, owners and operators slashed costs, put assets up for sale, or turned them over to lenders. Some plants were even subjugated to non-dispatch status, mothballed mere years after completing construction. For many merchant operators, the painful question that kept surfacing was, “Now what?”
The answer may lie in a previously overlooked number used to dispatch these combined (and simple)-cycle units. Despite increasing sophistication in market models, technical and financial instruments, and contractual structures, very little attention had been focused on the effect that O&M costs have on dispatch decisions. In many cases, the O&M number has been applied as just a static plug into an equation, the assumption being that it won’t affect the final profitability enough to cause concern.
In a world of baseload units and ever-declining natural gas prices, this may have held true. But with today’s narrow spark spreads and the need to squeeze profits from a brutally competitive market, the O&M number warrants an increasingly important role in the decision making process.
Consider a merchant plant selling into the New England power market. Given some reasonable assumptions as to the plant’s operating characteristics (two F-class turbines, 750 MW capacity, 7000 Btu/kWh), a general dispatch model can be constructed using the O&M cost accounting methods currently being used in the marketplace. Figure 1 provides a summary of the hypothetical plant’s operations. However, using a static plug of O&M costs in the model (adjusted once a year, as is the procedure of most operators) fails to fully capture this plant’s inherent value. Dynamically optimizing the O&M cost calculation to account for changing market conditions and maintenance schedules can substantially increase the plant’s value, defined here as the net present value of the plant’s profit. In this particular case, the plant’s value can be increased by up to 13 percent.
Why the dramatic difference? The reason lies in the complex, ever-changing maintenance costs of the turbines. A significant portion (nearly 60 percent in many instances) of a combined-cycle plant’s non-fuel O&M cost is tied to the combustion turbines’ outage expenses.
A combustion turbine has three main outages during its lifetime: a combustion inspection (CI) to review the combustor section of the turbine; a hot gas path inspection (HGP) to review both the combustor and all of the downstream components; and a major inspection (MI) that considers the entire rotor, including the cold compressor section. Because the scope of work increases with each successive level of inspection, the outage expense and downtime associated with the inspection also increases. An MI, for example, may require the turbine to be down for nearly a month.
Complicating matters further is that outage scheduling depends heavily upon the turbine’s running regime. Turbine maintenance is triggered by one of two metrics: hours or starts. For example, a typical turbine may be limited to 8,000 hours or 400 starts before requiring an outage, much in the same way that an automobile requires an oil change every three months or 3,000 miles, whichever comes first. Should the turbine operate under an hours-limited regime, the outage schedule may look far different than that for a turbine operating under a starts-limited regime. Consider the turbines in Figure 2, each of which operates under a different maintenance regime. The one operating under a starts-limited maintenance regime has not only exchanged a relatively low-cost CI for a more expensive HGP, but the timing of the first HGP has also been moved up. The result is a far more costly overall outage expense. Though every turbine model is slightly different in determining how hours-limited or starts-limited regimes look, most are subject to this increased expense if shifted to starts-limited maintenance.
For starts-limited operators, switching their turbines back into an hours-limited maintenance regime can yield substantial value. Optimizing O&M costs in the dispatch decision may be a strong beginning. Consider an operator who has amortized the starts-limited costs over the number of starts expected to be accrued on the machine. For instance, a turbine expected to undergo $24 million in outage expenses over 2,400 starts would equate a start-up “cost” of $10,000 per start. This number is passed along to the trader, who dutifully uses it in calculations to determine when to dispatch the unit.
However, if the trader were able to optimize the dispatch decision based on a more dynamic understanding of O&M costs, the trader may actually dispatch the unit more, or over more optimal revenue generating periods. By running the turbine more often, with less starts, the hours limit may become the constraint factor in determining the schedule of the next outage, and shift the turbine maintenance regime from that of a starts-limited regime to an hours-limited regime. Not only would this enable a switch from an “O&M cost per start” to an “O&M cost per hour,” but the estimated $24 million in expenses would also drop. In essence, dispatching the plant more (and more appropriately) lowers the per-start and per-hour cost of the facility.
Many operators attempt to achieve this improvement by applying both an hours-limited variable O&M (VOM) and a starts-limited “start charge.” But this method, too, can lead to suboptimal facility dispatch. Figure 3 profiles the hypothetical New England plant, where the original methodology used a $2.00/MWh VOM charge and a $12,000 start-up charge. These charges act as hurdles in the dispatching equation; if electricity prices are greater than the combination of the fuel costs and the dual O&M charges, then dispatch the unit. Since the operator will not know the actual costs incurred until some time after the plant has been dispatched, these charge numbers are actually more of an educated estimate. And, as Figure 3 indicates, the final analysis resulted in a $1.97/MWh VOM cost, closely matching the actual accrued expense.
The right-hand bars in Figure 3 show how an optimized routine does not do as good of a job at estimating the costs that will be incurred nor does it minimize the actual maintenance costs. However, the optimized routine does a much better job in maximizing the value of the facility. Figure 3 indicates that the optimized routine has slightly increased the VOM cost to $2.06/MWh, a situation that the operator may consider suboptimal. After all, maintenance costs are lower by $0.09/MWh (or nearly 5 percent) under the original methodology. However, minimizing maintenance cost is not the goal of the optimization routine. Rather, by focusing on maximizing profits, the optimization engine has run the turbine more often in early periods, has moved outages that would have occurred far in the future up (and hence has taken the penalty for the net present value effects of such), and is accumulating the benefit of running over marginal periods of time that were previously off-limits to the turbine. The result is a 5 percent increase in costs, but a 13 percent increase in profits.
It is the application of a “hurdle” rather than a “cost” that needs to be considered here. Most dispatching decisions are made under the assumption that the number being plugged into the dispatch calculation is a “variable cost.” For the New England plant, the estimated “cost” of $2.00/MWh is very close to the realized cost of $1.97/MWh. However, the quirk of the accounting method is such that a future, fixed cost is being applied in today’s terms as a variable cost. By submitting a static O&M charge number into the dispatch equation, the traders have, in essence, created a self-fulfilling prediction.
This methodology places a premium on accurate prediction versus an emphasis on profit maximization. An optimization model focuses instead on providing a “hurdle” target that maximizes the profits of the plant. The “hurdle” isn’t a true representation of cost, but rather, a dynamically changing mechanism that enables the trader to sell power into the marketplace during the most appropriate periods. Disconnecting this hurdle mechanism from a true accounting of costs is the first step that operating companies need to understand. For instance, the optimized New England plant example averages a VOM hurdle of $1.34/MWh while realizing a true “cost” of $2.06/MWh. The prediction is way off, but plant profit has increased substantially.
These two factors require a rethinking of O&M costs:
- Static costs vs. dynamic costs modeling:
How “static” is the O&M number being used? Can it adjust to changes in market conditions and operating profiles, or are significant operating opportunities being lost because of the inflexibility of this number?
- O&M dispatch hurdles vs. O&M accounting costs:
Is the O&M number designed to accurately predict the true “cost,” or is it a “hurdle” designed to maximize the plant’s profitability?
The impact on plant profitability can be truly significant. PowerAdvocate and CERA’s research has indicated that competitive plants may realize a 6 to 24 percent increase in asset value by optimizing their O&M costs. This range takes into account different turbine models, different predictive market scenarios, and different regional markets. In every case, a plant using a methodology of static hours-limited or starts-limited “costs” fails to optimize the plant’s value.
Paul Schuster is a director at PowerAdvocate (www.poweradvocate.com), a leading supplier of operations and supply chain solutions to the power industry. Through a combination of software, data, and services, PowerAdvocate helps clients increase asset values, drive fleetwide improvements, and reduce costs in a complex and ever-changing environment.
Steven Taub is research director for Cambridge Energy Research Associates’ (CERA) Emerging Generation Technologies Advisory Service. CERA (www.cera.com), a wholly owned subsidiary of IHS Energy, is a leading advisor to international energy companies, governments, financial institutions, and technology providers. CERA delivers critical knowledge and independent analysis on energy markets, geopolitics, industry trends, and strategy in all major energy sectors-oil and refined products, natural gas, and electric power-on a global and regional basis.
Cost Optimization in Practice
In late 2004, a large utility company retained PowerAdvocate and CERA to assess the opportunity associated with more sophisticated treatment of turbine O&M costs in dispatching one of their combined-cycle units. The competitive market had squeezed spark spreads and the facility was clearly feeling the pain of having to cut fixed and variable costs to compete effectively in the constrained market.
The plant has two F-class turbines operating in a 2 on 1 configuration with maximum capacity at around 500 MW. In addition, the facility has an actual variable cost from non-maintenance activities of $1.00/MWh. The facility was on a starts-limited maintenance regime and was using a static, annually updated start-up charge number in its dispatching decision. Using this simple treatment of O&M costs and CERA’s long-term gas and power pricing scenarios yielded a plant valuation of $46/kW.
PowerAdvocate and CERA applied the optimization model over a number of predictive market scenarios for the facility, iterating over each period a different start-up hurdle, a shutdown hurdle, and a VOM number. The objective of the optimization program was to maximize the NPV of cash flows associated with the plant. As each period progressed, the model looked back at the decisions made, looked ahead at how market conditions had changed since its previous calculations, and adjusted the hurdle numbers appropriately.
The results were very promising: with the optimized treatment of O&M, the NPV of the plant’s cash flows increased by 23 percent while shifting the turbine from a starts-limited maintenance regime to that of an hours-limited regime. Maintenance costs actually increased by 16 percent (from $10.2 million to $12.1 million) due to operating the facility more and the NPV effects of moving outages forward, but revenues increased by $34 million, or nearly 10 percent (from $322.9 million to $357.1 million). This was more than sufficient to offset the increased maintenance and hourly non-maintenance costs.